while searching and messing around I came across this file :
[url]http://www.clear.rice.edu/comp422/resources/cuda/pdf/cuobjdump.pdf[/url]
inside the doc/pdf of cuda installation.
It contains assembly instructions of Fermi and Kepler architectures. But without any explanation.
Apart from the usual Add, Mul, etc. Is there any other document or any thing that can explain the instructions? not how to code them, just their operation and what they do.
See this: [url]CUDA Toolkit Documentation and this if you want to use them inline w/ C code: [url]http://www.clear.rice.edu/comp422/resources/cuda/pdf/Using_Inline_PTX_Assembly_In_CUDA.pdf[/url]
The first link is out of date.
The second link works.
The third link is also out of date.
The only link that works is
http://docs.nvidia.com/cuda/pdf/ptx_isa_3.1.pdf and this redirects to CUDA Toolkit Documentation, the main CUDA Toolkit documentation page.
I did find a perl implementation of a Kepler assembler on gitHub: